1. Field of the Invention
The present invention relates generally to broadband network management, and more particularly to management of asynchronous transfer mode networks.
2. Background of the Invention
Traditionally, telecommunications service providers have offered basic services such as local and long distance exchange services for voice communications. More recently, with the explosive growth of the Internet and other data services, telecommunications service providers have expanded beyond basic telephone services to the provision of very high bandwidth network services. Examples of such newer network services include, for example, digital subscriber line (DSL), asymmetric digital subscriber line (ADSL), integrated services digital network (ISDN) digital subscriber line (IDSL), and the like. The underlying transmission facility supporting such high bandwidth networks may comprise an asynchronous transfer mode (ATM) network as shown in FIG. 1A.
ATM network or “cloud” 10 in FIG. 1A comprises one or more ATM switches 11-15. The switches may be interconnected in any suitable manner to provide redundancy in the network and to ensure high speed transmission of network packets. ATM switches are well-known in the art and are available from numerous switch vendors, including, for example, Lucent, Seimens and Northern Telecommunications. ATM switches typically comprise a processor, a memory and a backplane for supporting multiple network cards in a plurality of slots 16 as shown in FIG. 1B. A network card, for example, network card 17, typically comprises multiple ports 18 for supporting multiple communications paths. A distinct physical transmission cable, for example, a coaxial cable or fiber optic cable, may be connected to any given port on the network card. Each physical transmission cable carries thousands of logical circuits providing voice and data service.
Network service providers (NSP) 20-22 are provided connectivity to their end-users, subscribers 30, via ATM cloud 10 and central office DSLAM 40 via a permanent virtual circuit (PVC). The network traffic for each NSP is carried via a plurality of PVCs. A PVC is a connection that is established from a source end point to a destination end point without the ability of either end point to dynamically establish or release the connection. PVCs are manually implemented and must be manually released. A PVC is a “permanent” circuit because each PVC defines an end-to-end path for routing packets. However, the PVC is a “virtual” circuit because bandwidth from the cloud is utilized only when it is required. Each PVC is assigned a virtual path identifier (VPI) and a virtual connection identifier (VCI), which together identify the virtual circuit's end points, in accordance with well-known ATM standard specifications. Each ATM packet transmitted in ATM network 10 includes a VPI field and a VCI field in a packet's header. Within ATM network 10, the combination of VPI/VCI must be unique for each PVC at the network interface point (the source end and destination end of the PVC) to the ATM network, i.e., the physical network interface.
As described above, each ATM switch 11-15 supports multiple network cards, and each network card supports multiple physical connections. However, in conventional ATM networks, the lack of tools for adequately planning capacity changes in the network has proven to be a problem. For example, if every slot 16 on ATM switch 11 is configured with a network card supporting a eight ports (i.e., physical connections), and each port was configured with over one thousand PVCs, an unacceptable performance may result. One method for reducing such problems in conventional ATM networks has been to initially under-build the ATM network. That is, for example, an ATM switch having twelve slots may only be filled with ten cards, and each card may only have connections going into six of the eight available ports.
While an ATM network capacity manager may have anticipated such limitations when the network was designed, an accurate capacity plan may not be achievable without real-world network traffic being observed. This may be particularly true in situations where the demand for such high-bandwidth network services could not be adequately predicted. After an ATM network has been established, any changes to the physical card and port configuration require re-provisioning of every PVC affected. Accordingly, even moving one ATM physical connection from one switch to another may require re-provisioning of thousands of PVCs, which means hours and hours of service interruption.
Because each PVC must have a unique VPI/VCI at each network interface (it takes two network interfaces to create a destination and end point), the conventional method for re-provisioning a PVC required the following general steps:
1. select the new interface port;
2. unplug physical connection from old port and terminate to the new port; and
3. reassign, or relocate, the logical or virtual circuits to the new port.
This methodology is necessary due to the fact that the stationary end of the virtual circuit (the side not being relocated) cannot be manipulated due to the restriction of the unique VPI/VCI for the interface port.
The problem with implementing these steps in a conventional manner is that it may take several hours or even days to complete, resulting in unacceptable network downtime leading to customer service impairment. For example, an existing network may be configured as shown in the FIG. 2A. In this example, multiple PVCs are assigned on the DSLAM end to port two on the card in slot five (denoted “C5/P2”) of ATM switch 200. The NSP end of these PVC's are assigned to port three of the card in slot eight (C8/P3), port four of the card in slot 6 (C6/P4), and port two of the card in slot 8 (C8/P2) on ATM switch 200. If the ATM network manager needs to move all of these PVCs terminated to the DSLAM through the circuit connected to C5/P2 208 to a different card, for example, to port four on the card in slot seven (C7/P4), the ATM manager must manually reassign each virtual circuit associated with 208 manually as follows:
1. As shown in FIG. 2B, each PVC must be manually relocated to the port C7/P4. For example, PVC 202 is originally assigned a connection between C5/P2 and C8/P3 202, as shown in FIG. 2A. At this point, the new PVC, PVC 204 is an inactive PVC that will not carry traffic on the ATM network. Since the DSLAM 206 has not been connected to the new circuit 209, the PVC is inactive and will not be available to the end-users. This step typically takes about three minutes to perform for each PVC. Accordingly, if there are one thousand affected PVCs, the total downtime for this step alone will be fifty hours.
2. As shown in FIG. 2C, after the PVCs have been terminated to the new physical connection, circuit connection 209 is created by unplugging 208 from the old port. In other words, connection 208 is replaced by connection 209; and
3. As shown in FIG. 2D, the final step in the process is to restore service to each PVC after the physical circuit has been reconnected. At this point in the process, PVC 204 is active and available to NSP 20.
Step 2 is repeated for each PVC that has been moved. Again, this step requires about three minutes per PVC, resulting in a total downtime of fifty hours for this step and creating the unacceptable service condition requiring a different business process to complete the task.
As the above example illustrates, the conventional methods for moving network elements connected to an ATM cloud results in significant downtime for ATM network customers. A need therefore exists for a method of moving network elements with minimal network downtime for end-users associated with an ATM network.